Arlie's Journal

It turns out that I was wrong about Aisleriot lacking an auto-build feature. If you double click on a foundation pile, it will move any cards it can from tableau to any foundation. You can also apparently get the same effect by pressing the space bar with your mouse hovering over a foundation pile.

This is preferable to the online solitaire I used's habit of always auto-moving pretty much any card to a foundation if possible. That interferes with play, as you may want it left on the tabeau so as to build down on it. I've regularly had to unbuild foundation piles in that game, and sometimes had to use "undo" for the purpose, as there was nothing left on the tableau which could receive the cards. And if you moved a single card while in that undone state, without thereby making the cards ineligible for the foundation pile(s), they'd be immediately moved back.

What's not so great is how I found it. A misclick had triggered auto-build while I was playing. This told me the feature existed. I didn't know what i had done. There was no relevant menu item or icon, or explanation in the help menu. The man page tells me that you can doubleclick on a single card to move it to a foundation, but I saw nothing there about moving all eligible cards at once.

I finally found the information in a thread about an enhancement request. This functionality was not what the enhancement had asked for, but overlapped enough that someone mentioned it.

Per the man page, there's also "excellent Gnome documentation", which may be the same as the documentaiton quoted in said man page, complete with version number. But my search for this specific feature didn't provide a link to this documentation.

FWIW, the Gemini LLM bot in Google search agreed the functionality existed, but gave a description of how to use it that lacked the key detail that you should doubleclick on a foundation rather than on the game in general, or on the card to be moved. Possibly it was actually telling me what I already knew - that you can double-click on a single card in the tableau to automatically move it and only it to a foundation if it is eligible - having conflated two related questions.

[Update: the method involving hitting the space bar while the mouse points at a foundation does not work. But OTOH, the description just said one needed to "hover", not any particular place where the mouse should be hovering. I had guessed "over a foundation card" to match the doubleclick behaviour. So there may be some arcane recipe that actually works - or not. Bad documentation availability; no biscuit!]

I subscribe Richard Hanania's substack. He's a political blogger, included in my feed to provide one of a variety of viewpoints, without being massively unpleasant to read.

He's recently written a book, which he shills at every opportunity - once per post as a matter of policy, and IIRC moving earlier and earlier in the post; there's also a giant banner at the top center of his blog's home page, but it just gives the name of the book embedded in what appears to be a very large logo, so if you didn't know about the book already it would be meaning free.

Today he sent out a paying-subscriber-locked post (i.e. a spam, from the POV of a free subscriber) entitled Sixteen Fun Facts About Me. In what universe would learning more about this blogger as a person be such an enticing proposition that it would induce people to give him money?

FWIW, I think this "I'm important to everyone" tendency is increasing, though I haven't made a systematic study. Perhaps he's off his meds. Perhaps he should be on some. And perhaps he's just in a bubble where no humbling feedback ever arrives.

It's time for me to start paying attention to his useful post to spam and/or drek ratio.

I then tried bypassing the initialize-from-prior-device, and doing things the slow way.

That has so far worked decently, though it's really silly that I have to download everything from my phone via the internet. First backup the old phone to Google's cloud, then restore from the backup. I imagine it will be slow. Fortunately, I can use the device while it's running.

Meanwhile, none of my settings appear to have carried over (so far), and the phone still doesn't wake when moved. I was forced to create a PIN; the old phone operated password-free, which is probably unwise but saves a lot of time when using it for a clock, calculator, or similar app not requiring security. (It's beyond me why the morons who design these things couldn't put the security on the apps, not the phone. But they did the stupid thing, and now face ID is available as a workaround.)

I haven't put a sim in it, but by the looks of my carrier's web site, I can probably create an e-sim online and install it myself, without needing to visit their retail establishment. I won't put a sim in till it's otherwise working to my satisfaction, as that will commit me.

---

The pre-unlock screen is ugly, with a square containing 4 digits: 07 above 16. This might be the current time (12 hr clock, no AM or PM above, minutes below). Or it might be a reference to some sports team, for all I can tell. One thing for sure; this is not a reasonable format for a time, except maybe on a watch-sized screen.

Once I'm back on, it appears that google wants to tell me the wonders of its AI. Having once visited Gemini on the web, with results that felt like being stalked, I don't want any. Sadly, I can't buy a 5th G phone without such crap; at least the pixel 10A is too low end to support all of it, or to support it on device.

Apparently if I press and hold power I now get Gemini(*), rather than options like shutdown, reboot, take screenshot, etc. I wonder where those moved to, or if I'm now supposed to do them by talking to Gemini. I wonder if its voice recognition is finally reliable enough to be worth using, without extensive training of the speaker. (An early experiment was so aversive I haven't tried again, except when forced to by a phone tree that won't let me answer by pressing keys.)

After some opportunity to reject various privacy violations, the phone went into the tank with "setting up your phone". At that point it was about 30 minutes since I'd set a pin.

Now it wants to teach me various "gestures". Apparently the "go back" button I knew and loved has been replaced with something complicated, hard to remember, and impossible to check when needed, since you can't get out of whatever screen you are stuck in without knowing how. I foresee much searching of google on desktops, to figure out how to exit apps and other basic things, not to mention going "home", whatever that means.

The going back gesture doesn't work in the set of screens that teach gestures. This education also assumes you know what e.g. "go home" means. Maybe the same thing that gesture did on an older Android? Who knows.

Because of this, I didn't learn the gestures. I did eventually manage to perform the last one to the satisfaction of the teaching app, which wouldn't restart to give me the full description or the supposed meaning. But I got out.

Now I need to "choose how to navigate". Except nope, it switched to "set up a sim". It will be so much fun to have a phone that randomly pops up tasks for me to do.

My carefully curated home screens didn't transfer. They probably would have if the device-to-device method had worked. This is going to take work. But I've now got "settings" on home screen #1 - a major vistory.


--

(*) I tried pressing and holding power when frantically trying to get out of the gesture-teaching roach motel. At *that* time I got the usual screen, minus one option. Now I'm getting an "ask gemini" bar. *sigh*

Some years ago, when my previous physical linux box died, I found myself without my usual virtual solitaire program, which had been Aisleriot, only available for Linux.

While waiting for my replacement system to arrive, limited to only Mac and Android, I found https://online-solitaire.com/. It is a fairly well written solitaire app, originally focused on Klondike. Its monetization strategy is to display advertisements; my ad blocking extension in Safari took care of that. It would nag me every once in a while about how I should "support" them by viewing whatever they'd been paid to shill, but the nags were far less intrusive than the ads themselves would have been.

It's been working hard to force users to disable their adblocking software. When it became unusable on Safari with the ad blocker I had there, I considered installing Ad nauseam, an extension which doesn't display ads, but does click on every last one of them on the user's behalf. If I could configure it to only activate for selected sites, I could continue to use apps like this, and the sites themselves would probably not be motivated to investigate an increase in their revenues. (Their customers, who supply the ads, wouldn't be happy, but that's the point of ad nauseam.)

I decided that this was too dishonest for me. So I first stopped playing in Safari, with UBlock Origin continuing to work on Firefox. Then they figured out how to make their site equally unusable with UBlock Origin, and I took my toys and want home - to Aisleriot.

As always, there's a certain feeling of comfort in returning to an old familiar user interface, and my fingers easily retrained themselves. But the web page app had some features I do miss. You could tell it to give you deals that were guaranteed winnable. And when the game was essentially won, you could tell it to autoplay, saving you the trouble of manually unwinding tableau piles onto foundations, which Aisleriot requires if you want the win to register in the statistic.

OTOH, I don't miss other more negative features, beyond the ads, which could not be disabled.

Still it's a tiny loss I'd rather not have.

We survived PG&E's maintenance. After telling us power would be out from 9-5, they actually restored power around 11 AM, and texted me to tell me the work was complete at 11:23. By that time I was asleep, having gotten up far too early to deal with somewhat paranoid outage-prep.

I woke up again around 2 PM, saw the message, and brought up one of my computers. Lots of email while my systems were down, so I got out my metaphorical shovel.

At approximately 11:15, PG&E had sent me yet another spam email shaming me about my "high" power use. When I saw it, I clicked unsubscribe, which historically hasn't worked, and then hand constructed multiple filter rules to send these straight to spam, without losing things like messages about billing, or notifications of planned outages. They'll doubtless eventually change the subject line they use for these spams, but I should have piece for a while, and I do check my spam folder periodically, so will most likely notice if the rules get things they shouldn't.

Then I brought up the linux box. Because of the broken session management, reboots generate unwanted work, sending restored windows to the virtual desktop they came from. So once the box was up, I skipped the windows sorting task and went straight to installing all pending updates. Once installed, I rebooted again, to make sure that I was running the newly installed software.

I'm now procrastinating about sorting out (firefox) windows. This has given me time to remember that I have some more system administration work planned for this system, which should probably be followed by a reboot. I'll do that shortly, once I fetch another caffeinated beverage.

Talking of caffeinated beverages, i went a little wild before the outage, filling two small thermoses and two insulated travel mugs. The travel mugs were unable to keep coffee warm from 8:30ish to 14:00 ish; after I woke up I poured the contents into nukable mugs and reheated them. Probably their limit is more like 2 than 6 hours.

The thermoses are known to keep tea warm for most of a full day of bridge, but not the following night; I'll tap one of them for my next hot drink.

Back in late April I was looking for workarounds for my too-many-browser-windows-and-tabs problem.
I posted https://arlie.dreamwidth.org/534793.html

Then life happened, and this was no longer near the top of my priority list.

Yesterday my Kubuntu system hung for a considerable time, looking for processes to kill because it was almost out of "memory". The true root cause of that was likely a configuration issue involving inadequate swap space provided by Kubuntu's automatic system configuration. I still need to research how to fix that, and how to report the problem with the installer.

But meanwhile I had other reasons to want to reduce my browser tab and window load on Firefox. See item #2 in my Moving to Linux: The Bad post from late March.

Today I resurrected my research notes from April, did some more research, and installed OneTab into a rarely-used Firefox instance to try it out. It turned to have most of the features I wanted, but hadn't found in its documentation or its reviews. So I installed it in the heavily used Firefox on the Linux system, and then into Safari on the MacOS system.

This tool allows you to save open tabs' URLs into what amounts to a specialized bookmark section, closing the tabs (and reducing load on your system). You can later reopen them individually or in groups, optionally removing them from that bookmark section as you do so.

Both versions work similarly, but not identically. The Safari version is willing to give me the total number of tabs in each group, and overall, but not to show me the actual URLs; in fact, it's prone to giving pages a name of "Untitled" and an icon of a blank piece of paper. The Firefox version can't count, but it can both figure out better page titles and display their actual URLs.

After installing OneTab, I closed an awful lot of tabs into OneTab. (109 from Safari; probably a similar number on Linux' Firefox.)

I look forward hopefully to fewer linux hangs, and fewer random exits of an idle Safari. And restarting Firefox on linux will involve a lot less moving of Firefox windows to their appropriate virtual desktop.

In the course of my experiments with the MacOS installation of Firefox, which involved a certain amount of exiting and restarting, I was aggravated by Firefox opening with a pop-up telling me I should upgrade. It could be closed - i.e. the upgrade wasn't forced - but it was still a time waster. So I researched how to disable the warning, and managed to do so in spite of MacOS' unwillingness to let anything started from the shell modify the contents of the Firefox App's system directory. (I missed clicking on the ask-once-and-can't-get-back query about whether terminal should have permission to modify that directory.) So I created the needed .json file in /tmp, and used the Finder GUI to create the needed sub-directory and move the file into it.

Phew. Much accomplished, but I've developed a headache.

p.s. One of the alternatives I examined was Firefox' Tab Groups feature. I rate it useless for this purpose, and perhaps also for anything else.

Lingonaut was founded as a non-profit rival to Duolingo, as a language teaching system. I've been following them for a while, but too cheap to contribute financially, which would have made me eligible for early test versions.

They now have Android and iOS apps in open beta. You can get the Android one from the app store; probably also the iOS one but I didn't try that.

I downloaded it about an hour ago, and I'm now having pots of fun with their German class.

It is a beta. There are some problems with the software, and with the class-ware, and it took me some time to systematically decode the UI.

But basically it's great. It's been drilling me on pronunciation - distinguishing similar-sounding words. I think it's already improving my ability to handle certain distinctions I've previously only handled from context, if at all.

Best of all, the folks on their Discord respond positively to problem reports. There was IMO an issue with account creation, which I reported, and got a response of "Forwarded it to development team to look into that", which was perfect.

Then I remembered to be a bit neurotypical-friendly, and expressed thanks and compliments. That may have felt better for them - or OTOH, the person responding may be as neurodivergent as I am, and didn't notice the initial omission. Either way, I'm happy.

[Edited to add: I speak and read German well enough to usually get what I wanted in a German restaurant; not well enough to discuss or appreciate poetry or even manage a children's book without a dictionary at hand. I haven't yet tried them with a language I basically don't know, or gotten anywhere near German material that's completely new to me. Also, for the record, they asked about my existing knowledge of German before I started.]

I follow a substack blogger named David Friedman. He's local to me, and I've visited his house. One of my sisters also knows him in person from the SCA. So he gets more slack from us than most substackers, even though we regard large chunks of his politics as wrong-headed at best.

He's a mildly famous person, in SCA circles, one of their earliest fighters, possibly their first combat-chosen king, or the first in his kingdom. He's also the son of Milton Friedman. He even has his own wikipedia article.

He believes a lot of things common among libertarians, minarchists, and similar. Some of them strike my sister and I as absurd. Others strike us as being inconceivable without the privileged upbringing he experienced. But he's sometimes interesting, and functions in my blog feed as one of a handful of readable right wing writers.

One of his beliefs is that the ongoing climate change is more good than bad. His evidence looks cherry-picked to me, but then, some quantity of his opponents appear to me to be merely reciting talking points they don't understand. David at least has some understanding of what he's citing - unlike either them or some quantity of his allies.

( Read more... )

Yesterday I received a mildly strange email that I ignored until I had time to focus on it. The email was ostensibly an invitation to someone's "exclusive" party. I did not recognize their name.

Yes, some idjits will send out blastogram invites to their parties, and call them "exclusive". (My first thought was that the letter came from my bridge club's email list, and the party was celebrating a milestone in the bridge life of someone who never plays the same day/time as I do.) But farther examination strongly suggested a phishing attack intended to obtain victims' gmail name and password via a fake google login.

ProtonMail, bless them, had "report phishing" as one of the extra options on every email (accessed via a "hamburger" menu). After warning me that clicking "OK" would result in the decrypted message being forwarded to them, it also automatically performed all the actions of their "Move to spam" action, which includes blocking the sender.

Done and dusted.

I've been thinking about what LLM's might be able to do that would be useful for me. Given their known tendencies, they'd need to be spot checked, and the part of the data that was used for anything serious more than spot checked.

I came up with two candidates:
1) Save me work by generating the original list of available cell phones to prune down to the least bad option to replace my aging Pixel 3a.
2) Give me a table of daily US stock market index prices, by date, over the past several years, expressed in each of several currencies, with US dollar prices as a comparison. (I'm curious about the extent to which the "good" stock market remains dependent on the falling US dollar, with US stock markets in fact falling as measured in Canadian dollars, Euros, Yen, etc. A generally reliable blogger posted to this effect some weeks ago; I want to know whether the trend continues, and when it started.)

In both cases, I can get the raw data myself, tediously feed it into a spreadsheet, and do the relevant computations. (In the first case, a simple sort by ascending screen width.) In both cases, particularly the latter, it would take huge amounts of effort, and be a giant pain in the arse. I've done the cell phone list before, manually, and it took ages. The stock market index list would be intractable without significant programming, involving a customized data scraper for each site offering part of the needed data.

My bridge partner, who's never seen an LLM he doesn't like, and is a big enough user that he pays for it, was enthusiastic enough to try the first experiment on the spot, with his cell phone.

( Read more... )